In the current study, we found significant associations between low serum salivary amylase and high serum ketones, and between low serum pancreatic amylase and low RQs in healthy subjects. These associations were independent of relevant confounders including age, BMI, HbA1c, and eGFR. However, the proportion of salivary to total amylase was unlikely to relate with high serum ketones and low RQ, although the proportion of salivary amylase was correlated with the level of blood glucose at early time point after starch loading in our previous study .
Because subjects in this study are healthy non-obese young female non-smokers, current results indicate fundamental relationship among serum salivary and pancreatic amylase and metabolic indices such as blood ketones and RQ. As both high serum ketones and low RQs reflect the high combustion of lipids compared with carbohydrates [4,5,6], our current findings suggest that individuals with low serum salivary and pancreatic amylases may obtain their energy predominantly by lipid combustion (fatty acid oxidation), which is consistent with our previous study that showed an association between low total serum amylase and ketonuria in a heterogeneous population with a broad range of age (25–79 years) .
Notably, in our previous study , no significant inverse correlation between serum ketones and serum salivary and pancreatic amylases, which were assessed as continuous variables, was observed. However, significant associations between low serum salivary amylase and high serum ketones were observed in this study. This discrepancy may depend on the difference in statistical methods between nonparametric correlation tests in the previous study and logistic regression analysis in the current study, likely because serum ketones were highly skewed almost to an undetectable level (Fig. 1a, b). In contrast, the previous study showed a positive correlation between serum pancreatic amylase and RQ, which is consistent with the observed association between low pancreatic amylase and low RQs in this study. The degree of skewness of RQ values is mild compared with those of serum ketones (Fig. 1c), which may contribute to the similar RQ results of this study and the previous study.
It has been shown that reduced rates of fat oxidation, namely, high RQ, may contribute to the predisposition to obesity or weight gain [16,17,18]. However, the predisposition to fat accumulation is associated with high tissue sensitivity to insulin [19, 20]. Schutz showed in his study  that high RQ, low fat oxidation, and high insulin sensitivity (predictors) were observed in the dynamic phase, whereas low RQ, high fat oxidation, and insulin resistance (outcomes) were observed in the static phase (compensated state). In line with this, we have considered potential underlying mechanism for the current findings (Additional file 2: Figure S2), although this study is a cross-sectional study in nature. Baseline individual levels of serum amylases, which are genetically determined in most cases, are likely to be influenced by other factors including obesity and insulin resistance, eventually resulting in alternation of energy metabolism. The subjects with low serum salivary or pancreatic amylases observed in this study may be at the static phase or feedback phase because low RQ and high serum ketones (high fat oxidation) were associated with low serum pancreatic and salivary amylase, respectively.
Although high levels of ketones often reflect a deficiency in insulin secretion in diabetic patients [2,3,4], no subjects in this study had diabetes or impaired glucose metabolism. By contrast, the ketogenic diet, which involves carbohydrate restriction, frequently causes elevated serum ketones, even in healthy individuals [3, 4]. Because current subjects were young women aged 20-39 years, a proportion of the subjects may have conducted such a diet during this study, which may have contributed to the high levels of serum ketones, especially in the fasted state in the morning. This issue deserves further study.
Although current study indicates that fatty acids may be predominantly used as an energy source in individuals with low serum amylase, it is unclear if a diet rich in lipids is suitable for individuals with low serum amylase. Taken together, our studies suggest that a close relationship may exist between serum amylase and the specific type of macronutrient combustion used for energy.
While increases in serum amylase occur because of leakage from salivary glands and the pancreas, the clinical relevance of this remains unknown. Circulating amylases might just be a marker of leaks or damage, albeit several investigators have suggested a feedback system between serum amylase and insulin action [10, 11, 21]. Insulin resistance may downregulate the production of amylase , possibly for the purpose of reducing absorption of glucose digested from starch. On the contrary, high levels of serum amylase can reduce the secretion of insulin in the pancreas . However, it is unknown whether this plausible feedback system is also applicable in the salivary gland.
In conclusion, the current results obtained from the reanalysis of our previous study confirm the high combustion of lipids for energy in individuals with low serum amylases, suggesting a close relationship between circulating amylases and internal energy production.